F1 run_checks.sh exited 1 on any machine WITH SageMath. set -euo pipefail made
the declared-optional cross-checks fatal, so the script passed without the
optional tooling and failed with it. Optional calls are now guarded by
run_optional(), failures are counted and reported non-fatally, and the script
ends exit 0 with a summary line.
F2 false assertion in p28_lattice_hypotheses_certificate.sage. Component 1 read
216 + 108x + 46x; b2 = (1,2,1,0) contributes 216x, so the true value is
216 + 216x + 46x. The load-bearing identity below it (Bcomb == 4x*Crow) was
always correct, so no mathematics changes.
F3 Sage 10.9 raised 'keys do not match self's parent' at the subs() call: q
lives in the fraction field K while .numerator()/.denominator() return elements
of the underlying polynomial ring. The substitution key is now coerced into the
polynomial parent. This error had been MASKING F2.
Verified after the fixes:
run_checks.sh exit 0, 48 mandatory PASS (up from 43 -- the previously failing
certificate now runs to completion), 0 optional failures
falsifiability intact: 64R-44 -> 64R-43 exits 1, and 236337691420383 -> ...384
exits 1
pdflatex x3: 0 errors, 0 warnings, 0 undefined, 17 pages, 0 broken refs
Documentation: states plainly that no base is claimed superior to any other --
the non-injectivity holds for every b >= 2, [0,1] and [1] collide in decimal
exactly as in octal, and both repairs are stated for general b. Base 8 is only
the inherited worked example. Without this a reader could take the radix work
for a claim that base 8 beats base 10 or binary, which is not claimed anywhere.
Housekeeping: removes three Sage preparser .sage.py outputs that an earlier
'git add -A' in this branch had wrongly committed, and adds a .gitignore for
Sage and LaTeX build artifacts. Release zip and PDF rebuilt.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01WY6SfRYvm8zFKMX9GcjS8u
allaun
477e8ddb77